Is the plato formula better? Because just converting units shouldn't make it any more precise.

"Better" is kind of a loaded word. What Kai's saying is that when using (apparent) attenuation to estimate alcohol content, working in degrees Plato will give a more accurate result, because the units scale linearly with sugar content, by definition. SG, on the other hand, scales as a function of density.

So, for example, a 20°P wort contains twice as much sugar per unit volume as a 10°P wort. A 1.080 wort contains about 1.93 times as much sugar per unit volume as a 1.040 wort.